Models for Configuring Large-Scale Distributed Computing Systems
Bezalel Gavish · AT&T Technical Journal · 1985
This paper presents a model for designing the architecture of a distributed computing system of the type used to support the management and control activities of a large corporation. The input to the design process consists of two major data components describing the inputs and outputs to the information system and the relationships between them. Based on this information and the structure of communication and processing costs, an optimization model is formulated. It aggregates transactions in distributed databases, selects the locations in which those databases will be placed, assigns data sources to those databases, and selects for each report the report generation location. The problem is formulated as a combinatorial optimization problem and procedures are developed for computing lower bounds on the value of the optimal solution and heuristics for generating good feasible solutions for the problem. The procedures were tested on several examples and have generated good initial designs. Computational examples are presented to design problems including organizations with a hierarchical structure.